Research Of GPU-accelerated Rendering Method Of Map Symbols Concerning Visual Variable Similarity

The symbolization of various map elements is important in cartography research.And the quality and efficiency of symbolized rendering of spatial data has always been a difficult and hot issue in the research of geographical visualization. High quality rendering effect and more fluid user experience is the necessary demand of vector map service. Symbolic rendering efficiency and quality are important factors to affect the application of vector map. Therefore, it has important research value and practical significance to study the efficient symbolic rendering methods so that to reduce the response time of the system, and also to study the effect of high quality symbolic rendering so that to transfer more abundant semantic information and enhance the effect of symbolic rendering.With the development of graphics hardware technology, as well as the continuely enhancement of graphics processing and computing capabilities, the GPU-based rendering method has become a hot research topic in recent years. The graphics hardware acceleration technology is introduced into the symbolic rendering of the spatial data, which can make full use of the advantages of the fast rendering and flexible implementation of the programmable graphics pipeline, and then assist to solve the speed bottleneck of symbolic rendering. Compared with the software-based symbolic rendering methods, the GPU programmable graphics pipeline technology can provide hardware accelerated rendering performance for symbolic rendering of spatial data.However, traditional GPU-based methods for rendering spatial elements are mainly focused on the specific graphics rendering algorithm itself, the general flow of map symbol rendering is seldom considered in the same time. That has resulted in the difficulty to meet the demands of rendering various thematic mapsIn this paper, the visual variable similarity features of map symbol are considered as the starting point. Through the analysis and optimization of the traditional design of symbolic rendering process, two models have been studied: the primitive-based map symbol data model and the visual variable similarity driven symbolic process model.On this basis, with the help of GPU rendering advantages, the GPU-based method is introcued which take the visual similarity variable into consideration.The main research contents and results are structured as follows:(1) The primitive-based map symbol data model. The data model of map symbol is the basis of symbolic rendering. Firstly, the advantages and disadvantages of the existing map symbol data models are compared and analyzed. Combining the embodiment of the law of graphic elements and the ability of the Path method to describe the symbol graph, a primitive-based symbol data model is proposed. Visual variable is the basis of map symbol expression. In the symbolic rendering process,visual variable should be mapped into the attribute of symbolic elements and the corresponding drawing parameters as well. The qualitative description of the symbol visual variable is changed into quantitative attribute parameters, which is conducive to the judgment of visual variables.(2) The visual variable similarity driven symbolic process model. In the process of map symbol rendering, there are a lot of redundant operations which restrict the efficiency of symbolic rendering. To solve this problem, this paper presents a visual variable similarity driven symbolic model. By analyzing the similarity of visual variables between the same map symbols and different map symbols, this paper summarized the impacts on the reusability of various rendering sub processes. And a hierarchical cache strategy is designed to cross-share the rendering process,so that to achieve the effective elimination of redundant operation.(3) The GPU-based rendering method considering the visual variable similarity of map symbols. Based on the proposed primitive-based map symbol data model and the law of symbol primitives, the corresponding map symbol rendering method is designed,which employs the GPU rendering technologies. From the symbol I/O stage, the symbol element discrete stage, then to the symbol rasterization stage, the corresponding cache construction methods are designed and implementated. Meanwhile, to meet the basic requirements of GPU rendering, a triangulation method is designed to deal with the discrete points of linear elements; and the trapezoidal discrete method of area elements which based on horizontal scanning is designed. On these basis, combined with the construction of symbol cache (both in GPU and CPU), using the discrete results of spatial elements, the optimized drawing methods for point, line, area map symbols are introduced.(4) The construction of the prototype system for map visualization. On the basis of the first three research contents, this paper describes the system framework of the prototype system, and the optimized drawing method for point, line and area symbol are implemented. Through the comparison experiments with existing map symbol rendering methods (GDI/GDI+, AGG, Cairo), the feasibility and efficiency of the proposed method is verified. The optimized drawing method can effectively eliminate the redundant operation and improve the efficiency of symbol rendering.Considering the current development trend of map rendering and spatial information visualization technology,"The primitive-basde map symbol data model"and "The visual variable similarity driven symbolic process model" are proposed. With the advantage of hardware accelerated rendering, the visual variable similarity driven map symbol drawing method of is studied. Using this method can reduce the cost in symbolic rendering process, which is conducive to the improvement of the efficiency of symbolic rendering. Therefore, it can promote the two-dimensional spatial data and symbol resources applied into three-dimensional visualization system. Through the proposed method in this paper, it can provide new research technique for spatial information visualization. MeanWhile, it can provide a reference solution for fusing display of vector map on Virtual Geographical Scenes.